Which Holding Adds the Most Risk to My Portfolio?
The biggest position is often not the riskiest. Here is how risk contribution works, how to estimate it with public data, and a worked example where a 15% holding drives 30% of the swings.
Some risk pays you and some does not. Here is how to find the unpaid part first (duplicate funds, correlated clusters and single-stock bets), with real numbers on what each fix did historically.
What is your own portfolio exposed to? Check it with Exent →
Partly. You can remove risk that the market does not pay for without giving up expected return, but once that is gone, less risk usually means less return. The useful skill is telling the two apart.
Finance splits risk into two kinds. Market risk is the swing every stock shares with the economy; over long periods investors have been paid for bearing it. Stock-specific risk is what happens to one company alone; it can be diversified away, so the market does not pay a premium for carrying it. The systematic vs idiosyncratic risk entry explains the split. A practical order of work follows: duplicates first, then correlated clusters, then oversized single stocks, and only then the overall stock exposure.
Often, yes. Funds with different names can track nearly the same thing. Two holdings with a correlation near 1.00 add cost and complexity but no diversification. Exent’s data, 60 monthly returns to Sep 30, 2026:
| Pair | Correlation (5 years, monthly) | What it means |
|---|---|---|
| SPY and VOO | 1.00 | Same index (S&P 500), different issuer |
| SPY and VTI | 1.00 | Total US market is mostly the S&P 500 by weight |
| QQQ and VGT | 0.97 | Nasdaq-100 and a US tech fund moved almost as one |
| QQQ and IVW | 0.97 | Nasdaq-100 and S&P 500 Growth |
| QQQ and XLK | 0.96 | Nasdaq-100 and the tech sector fund |
| SPY and QQQ | 0.92 | Different index, same big companies at the top |
Merging duplicates does not change what you are exposed to. It just makes the real exposure visible, which is the precondition for every other step. Common pairs are listed in ETF overlap: common pairs.
Look for clusters: groups of holdings that rise and fall together, even if they are different securities. A cluster behaves like one large position. In a sample portfolio, everything sits in the same cluster, and one holding carries far more risk than its weight suggests.
| Sample portfolio | Weight | Correlation with QQQ | Estimated share of volatility |
|---|---|---|---|
| SPY (S&P 500 fund) | 40% | 0.92 | 29% |
| QQQ (Nasdaq-100 fund) | 25% | 1.00 | 23% |
| AAPL (Apple) | 20% | 0.69 | 18% |
| NVDA (Nvidia) | 15% | 0.79 | 30% |
Sample portfolio, same data as of Sep 30, 2026. Shares estimated with a one-factor model from each holding’s 5-year volatility and beta to the S&P 500; it ignores links beyond the market, so it understates how tightly tech names move together. By this estimate, 89% of the portfolio’s variance came from the stock market as a whole. The method is explained in which holding adds the most risk.
Mostly not, on the evidence. Of the 465 largest US stocks with five years of data, only 101 had a higher Sharpe ratio than SPY over the 5 years to Sep 30, 2026, and 183 had a higher total return (Exent’s data; the list is today’s largest companies, so it flatters past winners through survivorship bias). A concentrated stock can still be the right call, but the odds that its extra volatility is paid for are low.
Real portfolios carry a lot of it. In an anonymized sample of 144 portfolios uploaded by our users (internal accounts excluded), the median largest holding was 26.1% of the money and the median top five holdings were 77.2%. Trimming an oversized position toward a broad fund removes stock-specific risk while keeping market exposure. Whether it lowers return depends on that one stock, which is exactly the bet you are choosing to keep or not.
A holding with low correlation to the rest can lower total volatility even if it is volatile on its own. One direct test is the volatility of a portfolio that is 95% SPY and 5% the holding, over the same 5 years (SPY alone: 15.7%).
| Added 5% | Correlation with SPY | Its 5-year annual return | Volatility of the 95/5 mix |
|---|---|---|---|
| JNJ (Johnson & Johnson) | 0.19 | 13.4% | 15.1% |
| XOM (Exxon Mobil) | 0.13 | 26.8% | 15.1% |
| GLD (gold) | 0.17 | 18.3% | 15.1% |
| UUP (US dollar) | −0.43 | 5.7% | 14.7% |
| NVDA (Nvidia) | 0.69 | 61.7% | 16.7% |
| GBTC (bitcoin fund) | 0.50 | 13.8% | 16.9% |
| SPY (for reference) | 1.00 | 13.6% | 15.7% |
Over this window, some low-correlation holdings cut risk and had also returned as much as or more than the index. That is the closest thing to risk reduction without a return cost, but it is backward-looking: past returns of any single holding are not a forecast. The assets negatively correlated with stocks guide covers how these behaved in crashes.
Cutting stock market exposure itself. Our backtests, monthly data from July 2016 to September 2026 (hypothetical, before costs and taxes):
| Backtest | Annual return | Volatility | Sharpe | Max drawdown |
|---|---|---|---|---|
| S&P 500 (SPY) | 15.2% | 15.2% | 0.88 | 23.9% |
| 60/40 SPY/AGG, monthly rebalanced | 9.5% | 10.1% | 0.75 | 20.0% |
| Minimum volatility (USMV) | 9.4% | 12.2% | 0.63 | 19.1% |
| 10-month trend filter to T-bills | 9.4% | 11.6% | 0.66 | 21.9% |
Each cut volatility by a third or more but gave up more return than risk, so the Sharpe ratio fell. In this decade bonds were a weak partner (they lost money in 2022), and that may not repeat. Details on the 60/40 and minimum volatility backtest pages.
More methods sit under portfolio optimization.
Diversification is the closest one: removing stock-specific risk that the market does not pay for can lower volatility without lowering expected return. Beyond that, less risk has usually meant less return.
Not if they track overlapping indexes. In Exent’s data SPY, VOO and VTI had a correlation of 1.00 over five years, so holding all three adds no diversification.
Over long periods bonds have returned less than stocks, so they usually lower return as well as risk. From July 2016 to September 2026 a 60/40 mix returned 9.5% a year versus 15.2% for the S&P 500.
Trimming the holding with the largest share of risk relative to its weight usually lowers volatility the most per dollar moved. That is often the most volatile stock, not the biggest position.
Yes. In Exent’s data a 5% position in Johnson & Johnson or Exxon Mobil lowered an S&P 500 portfolio’s five-year volatility from 15.7% to 15.1%, because they moved differently from the index.
Upload a screenshot of your holdings and get concentration, overlap, crash exposure and a risk-matched benchmark in about a minute.
Educational content, not investment advice. Examples are illustrative; past performance does not guarantee future results.